This series is based on linux-next/akpm-base.
The series is split up like so:
- Patch 1 is a simple fixup which we should take in any case (even by itself).
- Patches 2-4 add the feature, basic support for it to the selftest, and docs.
- Patches 5-6 make the selftest configurable, so you can test one or the other
instead of always both. If we decide this is overcomplicated, we could just
drop these two patches and take the rest of the series.
Changelog:
v2->v3:
- Rebased onto linux-next/akpm-base, in order to be based on top of the
run_vmtests.sh refactor which was merged previously.
- Picked up some Reviewed-by's.
- Fixed ioctl definition (_IO instead of _IOWR), and stopped using
compat_ptr_ioctl since it is unneeded for ioctls which don't take a pointer.
- Removed the "handle_kernel_faults" bool, simplifying the code. The result is
logically equivalent, but simpler.
- Fixed userfaultfd selftest so it returns KSFT_SKIP appropriately.
- Reworded documentation per Shuah's feedback on v2.
- Improved example usage for userfaultfd selftest.
v1->v2:
- Add documentation update.
- Test *both* userfaultfd(2) and /dev/userfaultfd via the selftest.
Axel Rasmussen (6):
selftests: vm: add hugetlb_shared userfaultfd test to run_vmtests.sh
userfaultfd: add /dev/userfaultfd for fine grained access control
userfaultfd: selftests: modify selftest to use /dev/userfaultfd
userfaultfd: update documentation to describe /dev/userfaultfd
userfaultfd: selftests: make /dev/userfaultfd testing configurable
selftests: vm: add /dev/userfaultfd test cases to run_vmtests.sh
Documentation/admin-guide/mm/userfaultfd.rst | 40 ++++++++++-
Documentation/admin-guide/sysctl/vm.rst | 3 +
fs/userfaultfd.c | 76 ++++++++++++++++----
include/uapi/linux/userfaultfd.h | 4 ++
tools/testing/selftests/vm/run_vmtests.sh | 11 ++-
tools/testing/selftests/vm/userfaultfd.c | 66 ++++++++++++++---
6 files changed, 172 insertions(+), 28 deletions(-)
--
2.36.1.255.ge46751e96f-goog
Historically, it has been shown that intercepting kernel faults with
userfaultfd (thereby forcing the kernel to wait for an arbitrary amount
of time) can be exploited, or at least can make some kinds of exploits
easier. So, in 37cd0575b8 "userfaultfd: add UFFD_USER_MODE_ONLY" we
changed things so, in order for kernel faults to be handled by
userfaultfd, either the process needs CAP_SYS_PTRACE, or this sysctl
must be configured so that any unprivileged user can do it.
In a typical implementation of a hypervisor with live migration (take
QEMU/KVM as one such example), we do indeed need to be able to handle
kernel faults. But, both options above are less than ideal:
- Toggling the sysctl increases attack surface by allowing any
unprivileged user to do it.
- Granting the live migration process CAP_SYS_PTRACE gives it this
ability, but *also* the ability to "observe and control the
execution of another process [...], and examine and change [its]
memory and registers" (from ptrace(2)). This isn't something we need
or want to be able to do, so granting this permission violates the
"principle of least privilege".
This is all a long winded way to say: we want a more fine-grained way to
grant access to userfaultfd, without granting other additional
permissions at the same time.
To achieve this, add a /dev/userfaultfd misc device. This device
provides an alternative to the userfaultfd(2) syscall for the creation
of new userfaultfds. The idea is, any userfaultfds created this way will
be able to handle kernel faults, without the caller having any special
capabilities. Access to this mechanism is instead restricted using e.g.
standard filesystem permissions.
Signed-off-by: Axel Rasmussen <[email protected]>
---
fs/userfaultfd.c | 76 ++++++++++++++++++++++++++------
include/uapi/linux/userfaultfd.h | 4 ++
2 files changed, 66 insertions(+), 14 deletions(-)
diff --git a/fs/userfaultfd.c b/fs/userfaultfd.c
index e943370107d0..8b92c1398169 100644
--- a/fs/userfaultfd.c
+++ b/fs/userfaultfd.c
@@ -30,6 +30,7 @@
#include <linux/security.h>
#include <linux/hugetlb.h>
#include <linux/swapops.h>
+#include <linux/miscdevice.h>
int sysctl_unprivileged_userfaultfd __read_mostly;
@@ -413,13 +414,8 @@ vm_fault_t handle_userfault(struct vm_fault *vmf, unsigned long reason)
if (ctx->features & UFFD_FEATURE_SIGBUS)
goto out;
- if ((vmf->flags & FAULT_FLAG_USER) == 0 &&
- ctx->flags & UFFD_USER_MODE_ONLY) {
- printk_once(KERN_WARNING "uffd: Set unprivileged_userfaultfd "
- "sysctl knob to 1 if kernel faults must be handled "
- "without obtaining CAP_SYS_PTRACE capability\n");
+ if (!(vmf->flags & FAULT_FLAG_USER) && (ctx->flags & UFFD_USER_MODE_ONLY))
goto out;
- }
/*
* If it's already released don't get it. This avoids to loop
@@ -2052,19 +2048,33 @@ static void init_once_userfaultfd_ctx(void *mem)
seqcount_spinlock_init(&ctx->refile_seq, &ctx->fault_pending_wqh.lock);
}
-SYSCALL_DEFINE1(userfaultfd, int, flags)
+static inline bool userfaultfd_allowed(bool is_syscall, int flags)
+{
+ bool kernel_faults = !(flags & UFFD_USER_MODE_ONLY);
+ bool allow_unprivileged = sysctl_unprivileged_userfaultfd;
+
+ /* userfaultfd(2) access is controlled by sysctl + capability. */
+ if (is_syscall && kernel_faults) {
+ if (!allow_unprivileged && !capable(CAP_SYS_PTRACE))
+ return false;
+ }
+
+ /*
+ * For /dev/userfaultfd, access is to be controlled using e.g.
+ * permissions on the device node. We assume this is correctly
+ * configured by userspace, so we simply allow access here.
+ */
+
+ return true;
+}
+
+static int new_userfaultfd(bool is_syscall, int flags)
{
struct userfaultfd_ctx *ctx;
int fd;
- if (!sysctl_unprivileged_userfaultfd &&
- (flags & UFFD_USER_MODE_ONLY) == 0 &&
- !capable(CAP_SYS_PTRACE)) {
- printk_once(KERN_WARNING "uffd: Set unprivileged_userfaultfd "
- "sysctl knob to 1 if kernel faults must be handled "
- "without obtaining CAP_SYS_PTRACE capability\n");
+ if (!userfaultfd_allowed(is_syscall, flags))
return -EPERM;
- }
BUG_ON(!current->mm);
@@ -2083,6 +2093,10 @@ SYSCALL_DEFINE1(userfaultfd, int, flags)
refcount_set(&ctx->refcount, 1);
ctx->flags = flags;
ctx->features = 0;
+ /*
+ * If UFFD_USER_MODE_ONLY is not set, then userfaultfd_allowed() above
+ * decided that kernel faults were allowed and should be handled.
+ */
ctx->released = false;
atomic_set(&ctx->mmap_changing, 0);
ctx->mm = current->mm;
@@ -2098,8 +2112,42 @@ SYSCALL_DEFINE1(userfaultfd, int, flags)
return fd;
}
+SYSCALL_DEFINE1(userfaultfd, int, flags)
+{
+ return new_userfaultfd(true, flags);
+}
+
+static int userfaultfd_dev_open(struct inode *inode, struct file *file)
+{
+ return 0;
+}
+
+static long userfaultfd_dev_ioctl(struct file *file, unsigned int cmd, unsigned long flags)
+{
+ if (cmd != USERFAULTFD_IOC_NEW)
+ return -EINVAL;
+
+ return new_userfaultfd(false, flags);
+}
+
+static const struct file_operations userfaultfd_dev_fops = {
+ .open = userfaultfd_dev_open,
+ .unlocked_ioctl = userfaultfd_dev_ioctl,
+ .compat_ioctl = userfaultfd_dev_ioctl,
+ .owner = THIS_MODULE,
+ .llseek = noop_llseek,
+};
+
+static struct miscdevice userfaultfd_misc = {
+ .minor = MISC_DYNAMIC_MINOR,
+ .name = "userfaultfd",
+ .fops = &userfaultfd_dev_fops
+};
+
static int __init userfaultfd_init(void)
{
+ WARN_ON(misc_register(&userfaultfd_misc));
+
userfaultfd_ctx_cachep = kmem_cache_create("userfaultfd_ctx_cache",
sizeof(struct userfaultfd_ctx),
0,
diff --git a/include/uapi/linux/userfaultfd.h b/include/uapi/linux/userfaultfd.h
index 7d32b1e797fb..005e5e306266 100644
--- a/include/uapi/linux/userfaultfd.h
+++ b/include/uapi/linux/userfaultfd.h
@@ -12,6 +12,10 @@
#include <linux/types.h>
+/* ioctls for /dev/userfaultfd */
+#define USERFAULTFD_IOC 0xAA
+#define USERFAULTFD_IOC_NEW _IO(USERFAULTFD_IOC, 0x00)
+
/*
* If the UFFDIO_API is upgraded someday, the UFFDIO_UNREGISTER and
* UFFDIO_WAKE ioctls should be defined as _IOW and not as _IOR. In
--
2.36.1.255.ge46751e96f-goog
We clearly want to ensure both userfaultfd(2) and /dev/userfaultfd keep
working into the future, so just run the test twice, using each
interface.
Signed-off-by: Axel Rasmussen <[email protected]>
---
tools/testing/selftests/vm/userfaultfd.c | 37 +++++++++++++++++++++---
1 file changed, 33 insertions(+), 4 deletions(-)
diff --git a/tools/testing/selftests/vm/userfaultfd.c b/tools/testing/selftests/vm/userfaultfd.c
index 0bdfc1955229..1badb5d31bf9 100644
--- a/tools/testing/selftests/vm/userfaultfd.c
+++ b/tools/testing/selftests/vm/userfaultfd.c
@@ -77,6 +77,9 @@ static int bounces;
#define TEST_SHMEM 3
static int test_type;
+/* test using /dev/userfaultfd, instead of userfaultfd(2) */
+static bool test_dev_userfaultfd;
+
/* exercise the test_uffdio_*_eexist every ALARM_INTERVAL_SECS */
#define ALARM_INTERVAL_SECS 10
static volatile bool test_uffdio_copy_eexist = true;
@@ -154,12 +157,14 @@ static void usage(void)
ret, __LINE__); \
} while (0)
-#define err(fmt, ...) \
+#define errexit(exitcode, fmt, ...) \
do { \
_err(fmt, ##__VA_ARGS__); \
- exit(1); \
+ exit(exitcode); \
} while (0)
+#define err(fmt, ...) errexit(1, fmt, ##__VA_ARGS__)
+
static void uffd_stats_reset(struct uffd_stats *uffd_stats,
unsigned long n_cpus)
{
@@ -383,13 +388,31 @@ static void assert_expected_ioctls_present(uint64_t mode, uint64_t ioctls)
}
}
+static void __userfaultfd_open_dev(void)
+{
+ int fd;
+
+ uffd = -1;
+ fd = open("/dev/userfaultfd", O_RDWR | O_CLOEXEC);
+ if (fd < 0)
+ return;
+
+ uffd = ioctl(fd, USERFAULTFD_IOC_NEW,
+ O_CLOEXEC | O_NONBLOCK | UFFD_USER_MODE_ONLY);
+ close(fd);
+}
+
static void userfaultfd_open(uint64_t *features)
{
struct uffdio_api uffdio_api;
- uffd = syscall(__NR_userfaultfd, O_CLOEXEC | O_NONBLOCK | UFFD_USER_MODE_ONLY);
+ if (test_dev_userfaultfd)
+ __userfaultfd_open_dev();
+ else
+ uffd = syscall(__NR_userfaultfd,
+ O_CLOEXEC | O_NONBLOCK | UFFD_USER_MODE_ONLY);
if (uffd < 0)
- err("userfaultfd syscall not available in this kernel");
+ errexit(KSFT_SKIP, "creating userfaultfd failed");
uffd_flags = fcntl(uffd, F_GETFD, NULL);
uffdio_api.api = UFFD_API;
@@ -1691,6 +1714,12 @@ int main(int argc, char **argv)
}
printf("nr_pages: %lu, nr_pages_per_cpu: %lu\n",
nr_pages, nr_pages_per_cpu);
+
+ test_dev_userfaultfd = false;
+ if (userfaultfd_stress())
+ return 1;
+
+ test_dev_userfaultfd = true;
return userfaultfd_stress();
}
--
2.36.1.255.ge46751e96f-goog
This new mode was recently added to the userfaultfd selftest. We want to
exercise both userfaultfd(2) as well as /dev/userfaultfd, so add both
test cases to the script.
Reviewed-by: Shuah Khan <[email protected]>
Signed-off-by: Axel Rasmussen <[email protected]>
---
tools/testing/selftests/vm/run_vmtests.sh | 5 +++++
1 file changed, 5 insertions(+)
diff --git a/tools/testing/selftests/vm/run_vmtests.sh b/tools/testing/selftests/vm/run_vmtests.sh
index 7e102246dd9f..930c54eb5b4b 100755
--- a/tools/testing/selftests/vm/run_vmtests.sh
+++ b/tools/testing/selftests/vm/run_vmtests.sh
@@ -121,12 +121,17 @@ run_test ./gup_test -a
run_test ./gup_test -ct -F 0x1 0 19 0x1000
run_test ./userfaultfd anon 20 16
+run_test ./userfaultfd anon:dev 20 16
# Hugetlb tests require source and destination huge pages. Pass in half the
# size ($half_ufd_size_MB), which is used for *each*.
run_test ./userfaultfd hugetlb "$half_ufd_size_MB" 32
+run_test ./userfaultfd hugetlb:dev "$half_ufd_size_MB" 32
run_test ./userfaultfd hugetlb_shared "$half_ufd_size_MB" 32 "$mnt"/uffd-test
rm -f "$mnt"/uffd-test
+run_test ./userfaultfd hugetlb_shared:dev "$half_ufd_size_MB" 32 "$mnt"/uffd-test
+rm -f "$mnt"/uffd-test
run_test ./userfaultfd shmem 20 16
+run_test ./userfaultfd shmem:dev 20 16
#cleanup
umount "$mnt"
--
2.36.1.255.ge46751e96f-goog
On Wed, 1 Jun 2022 14:09:47 -0700 Axel Rasmussen <[email protected]> wrote:
> To achieve this, add a /dev/userfaultfd misc device. This device
> provides an alternative to the userfaultfd(2) syscall for the creation
> of new userfaultfds. The idea is, any userfaultfds created this way will
> be able to handle kernel faults, without the caller having any special
> capabilities. Access to this mechanism is instead restricted using e.g.
> standard filesystem permissions.
The use of a /dev node isn't pretty. Why can't this be done by
tweaking sys_userfaultfd() or by adding a sys_userfaultfd2()?
Peter, will you be completing review of this patchset?
On Mon, Jun 13, 2022 at 02:55:40PM -0700, Andrew Morton wrote:
> On Wed, 1 Jun 2022 14:09:47 -0700 Axel Rasmussen <[email protected]> wrote:
>
> > To achieve this, add a /dev/userfaultfd misc device. This device
> > provides an alternative to the userfaultfd(2) syscall for the creation
> > of new userfaultfds. The idea is, any userfaultfds created this way will
> > be able to handle kernel faults, without the caller having any special
> > capabilities. Access to this mechanism is instead restricted using e.g.
> > standard filesystem permissions.
>
> The use of a /dev node isn't pretty. Why can't this be done by
> tweaking sys_userfaultfd() or by adding a sys_userfaultfd2()?
>
> Peter, will you be completing review of this patchset?
Sorry to not have reviewed it proactively..
I think it's because I never had a good picture/understanding of what
should be the best security model for uffd, meanwhile I am (it seems) just
seeing more and more ways to "provide a safer uffd" by different people
using different ways.. and I never had time (and probably capability too..)
to figure out the correct approach if not to accept all options provided.
I think I'll just assume the whole thing is acked already from you
generally, then I'll read at least the implementation before the end of
tomorrow.
Thanks,
--
Peter Xu
On Mon, Jun 13, 2022 at 3:29 PM Peter Xu <[email protected]> wrote:
>
> On Mon, Jun 13, 2022 at 02:55:40PM -0700, Andrew Morton wrote:
> > On Wed, 1 Jun 2022 14:09:47 -0700 Axel Rasmussen <[email protected]> wrote:
> >
> > > To achieve this, add a /dev/userfaultfd misc device. This device
> > > provides an alternative to the userfaultfd(2) syscall for the creation
> > > of new userfaultfds. The idea is, any userfaultfds created this way will
> > > be able to handle kernel faults, without the caller having any special
> > > capabilities. Access to this mechanism is instead restricted using e.g.
> > > standard filesystem permissions.
> >
> > The use of a /dev node isn't pretty. Why can't this be done by
> > tweaking sys_userfaultfd() or by adding a sys_userfaultfd2()?
I think for any approach involving syscalls, we need to be able to
control access to who can call a syscall. Maybe there's another way
I'm not aware of, but I think today the only mechanism to do this is
capabilities. I proposed adding a CAP_USERFAULTFD for this purpose,
but that approach was rejected [1]. So, I'm not sure of another way
besides using a device node.
One thing that could potentially make this cleaner is, as one LWN
commenter pointed out, we could have open() on /dev/userfaultfd just
return a new userfaultfd directly, instead of this multi-step process
of open /dev/userfaultfd, NEW ioctl, then you get a userfaultfd. When
I wrote this originally it wasn't clear to me how to get that to
happen - open() doesn't directly return the result of our custom open
function pointer, as far as I can tell - but it could be investigated.
[1]: https://lore.kernel.org/lkml/[email protected]/T/
> >
> > Peter, will you be completing review of this patchset?
>
> Sorry to not have reviewed it proactively..
>
> I think it's because I never had a good picture/understanding of what
> should be the best security model for uffd, meanwhile I am (it seems) just
> seeing more and more ways to "provide a safer uffd" by different people
> using different ways.. and I never had time (and probably capability too..)
> to figure out the correct approach if not to accept all options provided.
Agreed, what we have right now is a bit of a mess of different
approaches. I think the reason for this is, there is no "perfect" way
to control access to features like this, so what we now have is
several different approaches with different tradeoffs.
From my perspective, the existing controls were simpler to implement,
but are not ideal because they require us to grant access to UFFD
*plus more stuff too*.
The approach I've proposed is the most granular, so it doesn't require
adding any extra permissions. But, I agree the interface is sort of
overcomplicated. :/ But, from my perspective, security in shared Cloud
computing environments where UFFD is used for live migration is
critical, so I prefer this tradeoff - I'll put up with a slightly
messier interface, if the gain is a very minimal set of privileges.
>
> I think I'll just assume the whole thing is acked already from you
> generally, then I'll read at least the implementation before the end of
> tomorrow.
>
> Thanks,
>
> --
> Peter Xu
>
Axel Rasmussen <[email protected]> writes:
> I think for any approach involving syscalls, we need to be able to
> control access to who can call a syscall. Maybe there's another way
> I'm not aware of, but I think today the only mechanism to do this is
> capabilities. I proposed adding a CAP_USERFAULTFD for this purpose,
> but that approach was rejected [1]. So, I'm not sure of another way
> besides using a device node.
I take it there's a reason why this can't be done with a security module
- either a custom module or a policy in one of the existing modules?
That sort of access control is just what security modules are supposed
to be for, after all.
Thanks,
jon
On Jun 13, 2022, at 3:38 PM, Axel Rasmussen <[email protected]> wrote:
> On Mon, Jun 13, 2022 at 3:29 PM Peter Xu <[email protected]> wrote:
>> On Mon, Jun 13, 2022 at 02:55:40PM -0700, Andrew Morton wrote:
>>> On Wed, 1 Jun 2022 14:09:47 -0700 Axel Rasmussen <[email protected]> wrote:
>>>
>>>> To achieve this, add a /dev/userfaultfd misc device. This device
>>>> provides an alternative to the userfaultfd(2) syscall for the creation
>>>> of new userfaultfds. The idea is, any userfaultfds created this way will
>>>> be able to handle kernel faults, without the caller having any special
>>>> capabilities. Access to this mechanism is instead restricted using e.g.
>>>> standard filesystem permissions.
>>>
>>> The use of a /dev node isn't pretty. Why can't this be done by
>>> tweaking sys_userfaultfd() or by adding a sys_userfaultfd2()?
>
> I think for any approach involving syscalls, we need to be able to
> control access to who can call a syscall. Maybe there's another way
> I'm not aware of, but I think today the only mechanism to do this is
> capabilities. I proposed adding a CAP_USERFAULTFD for this purpose,
> but that approach was rejected [1]. So, I'm not sure of another way
> besides using a device node.
>
> One thing that could potentially make this cleaner is, as one LWN
> commenter pointed out, we could have open() on /dev/userfaultfd just
> return a new userfaultfd directly, instead of this multi-step process
> of open /dev/userfaultfd, NEW ioctl, then you get a userfaultfd. When
> I wrote this originally it wasn't clear to me how to get that to
> happen - open() doesn't directly return the result of our custom open
> function pointer, as far as I can tell - but it could be investigated.
If this direction is pursued, I think that it would be better to set it as
/proc/[pid]/userfaultfd, which would allow remote monitors (processes) to
hook into userfaultfd of remote processes. I have a patch for that which
extends userfaultfd syscall, but /proc/[pid]/userfaultfd may be cleaner.
On Wed, Jun 01, 2022 at 02:09:47PM -0700, Axel Rasmussen wrote:
> Historically, it has been shown that intercepting kernel faults with
> userfaultfd (thereby forcing the kernel to wait for an arbitrary amount
> of time) can be exploited, or at least can make some kinds of exploits
> easier. So, in 37cd0575b8 "userfaultfd: add UFFD_USER_MODE_ONLY" we
> changed things so, in order for kernel faults to be handled by
> userfaultfd, either the process needs CAP_SYS_PTRACE, or this sysctl
> must be configured so that any unprivileged user can do it.
>
> In a typical implementation of a hypervisor with live migration (take
> QEMU/KVM as one such example), we do indeed need to be able to handle
> kernel faults. But, both options above are less than ideal:
>
> - Toggling the sysctl increases attack surface by allowing any
> unprivileged user to do it.
>
> - Granting the live migration process CAP_SYS_PTRACE gives it this
> ability, but *also* the ability to "observe and control the
> execution of another process [...], and examine and change [its]
> memory and registers" (from ptrace(2)). This isn't something we need
> or want to be able to do, so granting this permission violates the
> "principle of least privilege".
>
> This is all a long winded way to say: we want a more fine-grained way to
> grant access to userfaultfd, without granting other additional
> permissions at the same time.
>
> To achieve this, add a /dev/userfaultfd misc device. This device
> provides an alternative to the userfaultfd(2) syscall for the creation
> of new userfaultfds. The idea is, any userfaultfds created this way will
> be able to handle kernel faults, without the caller having any special
> capabilities. Access to this mechanism is instead restricted using e.g.
> standard filesystem permissions.
>
> Signed-off-by: Axel Rasmussen <[email protected]>
> ---
> fs/userfaultfd.c | 76 ++++++++++++++++++++++++++------
> include/uapi/linux/userfaultfd.h | 4 ++
> 2 files changed, 66 insertions(+), 14 deletions(-)
>
> diff --git a/fs/userfaultfd.c b/fs/userfaultfd.c
> index e943370107d0..8b92c1398169 100644
> --- a/fs/userfaultfd.c
> +++ b/fs/userfaultfd.c
> @@ -30,6 +30,7 @@
> #include <linux/security.h>
> #include <linux/hugetlb.h>
> #include <linux/swapops.h>
> +#include <linux/miscdevice.h>
>
> int sysctl_unprivileged_userfaultfd __read_mostly;
>
> @@ -413,13 +414,8 @@ vm_fault_t handle_userfault(struct vm_fault *vmf, unsigned long reason)
>
> if (ctx->features & UFFD_FEATURE_SIGBUS)
> goto out;
> - if ((vmf->flags & FAULT_FLAG_USER) == 0 &&
> - ctx->flags & UFFD_USER_MODE_ONLY) {
> - printk_once(KERN_WARNING "uffd: Set unprivileged_userfaultfd "
> - "sysctl knob to 1 if kernel faults must be handled "
> - "without obtaining CAP_SYS_PTRACE capability\n");
> + if (!(vmf->flags & FAULT_FLAG_USER) && (ctx->flags & UFFD_USER_MODE_ONLY))
> goto out;
> - }
>
> /*
> * If it's already released don't get it. This avoids to loop
> @@ -2052,19 +2048,33 @@ static void init_once_userfaultfd_ctx(void *mem)
> seqcount_spinlock_init(&ctx->refile_seq, &ctx->fault_pending_wqh.lock);
> }
>
> -SYSCALL_DEFINE1(userfaultfd, int, flags)
> +static inline bool userfaultfd_allowed(bool is_syscall, int flags)
> +{
> + bool kernel_faults = !(flags & UFFD_USER_MODE_ONLY);
> + bool allow_unprivileged = sysctl_unprivileged_userfaultfd;
> +
> + /* userfaultfd(2) access is controlled by sysctl + capability. */
> + if (is_syscall && kernel_faults) {
> + if (!allow_unprivileged && !capable(CAP_SYS_PTRACE))
> + return false;
> + }
> +
> + /*
> + * For /dev/userfaultfd, access is to be controlled using e.g.
> + * permissions on the device node. We assume this is correctly
> + * configured by userspace, so we simply allow access here.
> + */
> +
> + return true;
> +}
This helper reads a bit weird because potentially it constantly returns
"true" for !syscall use case but it's very not obvious..
Would it be cleaner to not pass in the bool at all? Something like (I also
un-nested some of the condition checks, hopefully it'll be easier to read):
bool userfaultfd_syscall_allowed(int flags)
{
/* Userspace-only page faults are always allowed */
if (flags & UFFD_USER_MODE_ONLY)
return true;
/*
* The user is requesting kernel fault capabilities. Privileged
* users are always allowed even for kernel fault traps.
*/
if (capable(CAP_SYS_PTRACE))
return true;
/* Whether we allow unprivileged users for kernel faults? */
return sysctl_unprivileged_userfaultfd;
}
Then below...
> +
> +static int new_userfaultfd(bool is_syscall, int flags)
> {
> struct userfaultfd_ctx *ctx;
> int fd;
>
> - if (!sysctl_unprivileged_userfaultfd &&
> - (flags & UFFD_USER_MODE_ONLY) == 0 &&
> - !capable(CAP_SYS_PTRACE)) {
> - printk_once(KERN_WARNING "uffd: Set unprivileged_userfaultfd "
> - "sysctl knob to 1 if kernel faults must be handled "
> - "without obtaining CAP_SYS_PTRACE capability\n");
> + if (!userfaultfd_allowed(is_syscall, flags))
> return -EPERM;
> - }
.. we could write it as:
if (is_syscall && !userfaultfd_syscall_allowed(flags))
return -EPERM;
What do you think?
>
> BUG_ON(!current->mm);
>
> @@ -2083,6 +2093,10 @@ SYSCALL_DEFINE1(userfaultfd, int, flags)
> refcount_set(&ctx->refcount, 1);
> ctx->flags = flags;
> ctx->features = 0;
> + /*
> + * If UFFD_USER_MODE_ONLY is not set, then userfaultfd_allowed() above
> + * decided that kernel faults were allowed and should be handled.
> + */
Hmm.. why this needs to be added above "released=false"? Did you want to
add this (perhaps) above "flags" instead?
IMHO when people reading the flags it'll be clear already on how it was
handled, the thing is the comment probably hide deep anyway so I'd consider
omitting it.
The rest looks good to me, thanks.
> ctx->released = false;
> atomic_set(&ctx->mmap_changing, 0);
> ctx->mm = current->mm;
> @@ -2098,8 +2112,42 @@ SYSCALL_DEFINE1(userfaultfd, int, flags)
> return fd;
> }
>
> +SYSCALL_DEFINE1(userfaultfd, int, flags)
> +{
> + return new_userfaultfd(true, flags);
> +}
> +
> +static int userfaultfd_dev_open(struct inode *inode, struct file *file)
> +{
> + return 0;
> +}
> +
> +static long userfaultfd_dev_ioctl(struct file *file, unsigned int cmd, unsigned long flags)
> +{
> + if (cmd != USERFAULTFD_IOC_NEW)
> + return -EINVAL;
> +
> + return new_userfaultfd(false, flags);
> +}
> +
> +static const struct file_operations userfaultfd_dev_fops = {
> + .open = userfaultfd_dev_open,
> + .unlocked_ioctl = userfaultfd_dev_ioctl,
> + .compat_ioctl = userfaultfd_dev_ioctl,
> + .owner = THIS_MODULE,
> + .llseek = noop_llseek,
> +};
> +
> +static struct miscdevice userfaultfd_misc = {
> + .minor = MISC_DYNAMIC_MINOR,
> + .name = "userfaultfd",
> + .fops = &userfaultfd_dev_fops
> +};
> +
> static int __init userfaultfd_init(void)
> {
> + WARN_ON(misc_register(&userfaultfd_misc));
> +
> userfaultfd_ctx_cachep = kmem_cache_create("userfaultfd_ctx_cache",
> sizeof(struct userfaultfd_ctx),
> 0,
> diff --git a/include/uapi/linux/userfaultfd.h b/include/uapi/linux/userfaultfd.h
> index 7d32b1e797fb..005e5e306266 100644
> --- a/include/uapi/linux/userfaultfd.h
> +++ b/include/uapi/linux/userfaultfd.h
> @@ -12,6 +12,10 @@
>
> #include <linux/types.h>
>
> +/* ioctls for /dev/userfaultfd */
> +#define USERFAULTFD_IOC 0xAA
> +#define USERFAULTFD_IOC_NEW _IO(USERFAULTFD_IOC, 0x00)
> +
> /*
> * If the UFFDIO_API is upgraded someday, the UFFDIO_UNREGISTER and
> * UFFDIO_WAKE ioctls should be defined as _IOW and not as _IOR. In
> --
> 2.36.1.255.ge46751e96f-goog
>
--
Peter Xu
On Wed, Jun 01, 2022 at 02:09:51PM -0700, Axel Rasmussen wrote:
> This new mode was recently added to the userfaultfd selftest. We want to
> exercise both userfaultfd(2) as well as /dev/userfaultfd, so add both
> test cases to the script.
>
> Reviewed-by: Shuah Khan <[email protected]>
> Signed-off-by: Axel Rasmussen <[email protected]>
Acked-by: Peter Xu <[email protected]>
--
Peter Xu
On Wed, Jun 01, 2022 at 02:09:48PM -0700, Axel Rasmussen wrote:
> We clearly want to ensure both userfaultfd(2) and /dev/userfaultfd keep
> working into the future, so just run the test twice, using each
> interface.
>
> Signed-off-by: Axel Rasmussen <[email protected]>
> ---
> tools/testing/selftests/vm/userfaultfd.c | 37 +++++++++++++++++++++---
> 1 file changed, 33 insertions(+), 4 deletions(-)
>
> diff --git a/tools/testing/selftests/vm/userfaultfd.c b/tools/testing/selftests/vm/userfaultfd.c
> index 0bdfc1955229..1badb5d31bf9 100644
> --- a/tools/testing/selftests/vm/userfaultfd.c
> +++ b/tools/testing/selftests/vm/userfaultfd.c
> @@ -77,6 +77,9 @@ static int bounces;
> #define TEST_SHMEM 3
> static int test_type;
>
> +/* test using /dev/userfaultfd, instead of userfaultfd(2) */
> +static bool test_dev_userfaultfd;
> +
> /* exercise the test_uffdio_*_eexist every ALARM_INTERVAL_SECS */
> #define ALARM_INTERVAL_SECS 10
> static volatile bool test_uffdio_copy_eexist = true;
> @@ -154,12 +157,14 @@ static void usage(void)
> ret, __LINE__); \
> } while (0)
>
> -#define err(fmt, ...) \
> +#define errexit(exitcode, fmt, ...) \
> do { \
> _err(fmt, ##__VA_ARGS__); \
> - exit(1); \
> + exit(exitcode); \
> } while (0)
>
> +#define err(fmt, ...) errexit(1, fmt, ##__VA_ARGS__)
> +
> static void uffd_stats_reset(struct uffd_stats *uffd_stats,
> unsigned long n_cpus)
> {
> @@ -383,13 +388,31 @@ static void assert_expected_ioctls_present(uint64_t mode, uint64_t ioctls)
> }
> }
>
> +static void __userfaultfd_open_dev(void)
> +{
> + int fd;
> +
> + uffd = -1;
> + fd = open("/dev/userfaultfd", O_RDWR | O_CLOEXEC);
> + if (fd < 0)
> + return;
> +
> + uffd = ioctl(fd, USERFAULTFD_IOC_NEW,
> + O_CLOEXEC | O_NONBLOCK | UFFD_USER_MODE_ONLY);
> + close(fd);
> +}
> +
> static void userfaultfd_open(uint64_t *features)
> {
> struct uffdio_api uffdio_api;
>
> - uffd = syscall(__NR_userfaultfd, O_CLOEXEC | O_NONBLOCK | UFFD_USER_MODE_ONLY);
> + if (test_dev_userfaultfd)
> + __userfaultfd_open_dev();
I can understand uffd is a global var, but still AFAICT that's trivially
easy to do the return to match the syscall case..
uffd = __userfaultfd_open_dev();
And since at it, it'll be great to make a macro:
#define UFFD_FLAGS (O_CLOEXEC | O_NONBLOCK | UFFD_USER_MODE_ONLY)
Thanks,
> + else
> + uffd = syscall(__NR_userfaultfd,
> + O_CLOEXEC | O_NONBLOCK | UFFD_USER_MODE_ONLY);
> if (uffd < 0)
> - err("userfaultfd syscall not available in this kernel");
> + errexit(KSFT_SKIP, "creating userfaultfd failed");
> uffd_flags = fcntl(uffd, F_GETFD, NULL);
>
> uffdio_api.api = UFFD_API;
> @@ -1691,6 +1714,12 @@ int main(int argc, char **argv)
> }
> printf("nr_pages: %lu, nr_pages_per_cpu: %lu\n",
> nr_pages, nr_pages_per_cpu);
> +
> + test_dev_userfaultfd = false;
> + if (userfaultfd_stress())
> + return 1;
> +
> + test_dev_userfaultfd = true;
> return userfaultfd_stress();
> }
>
> --
> 2.36.1.255.ge46751e96f-goog
>
--
Peter Xu
On Mon, Jun 13, 2022 at 4:23 PM Jonathan Corbet <[email protected]> wrote:
>
> Axel Rasmussen <[email protected]> writes:
>
> > I think for any approach involving syscalls, we need to be able to
> > control access to who can call a syscall. Maybe there's another way
> > I'm not aware of, but I think today the only mechanism to do this is
> > capabilities. I proposed adding a CAP_USERFAULTFD for this purpose,
> > but that approach was rejected [1]. So, I'm not sure of another way
> > besides using a device node.
>
> I take it there's a reason why this can't be done with a security module
> - either a custom module or a policy in one of the existing modules?
> That sort of access control is just what security modules are supposed
> to be for, after all.
>
> Thanks,
>
> jon
Admittedly I haven't tried proposing a patch, but I suspect there
would be pushback against adding an entirely new LSM just for this
case, similarly to the reasons the CAP_USERFAULTFD approach was
rejected.
For existing LSMs, I think SELinux can be used to restrict access to
syscalls. But then again, it's fairly heavy weight / difficult to
configure, and I suspect migrating production servers which don't use
it today would be a nontrivial undertaking. At least to me it seems
unfortunate to say, there isn't an obvious "safe" way to use
userfaultfd, without enabling + configuring selinux. (That assumes by
"safe" we mean, without granting wider-than necessary access to
userfaultfd, or without granting uffd-using processes more permissions
[root or CAP_SYS_PTRACE] to do their job.) I suspect if we do that
then in practice many? most? users will just either run UFFD programs
as root, or toggle the sysctl to allow unprivileged UFFD usage.
On Mon, Jun 13, 2022 at 5:10 PM Nadav Amit <[email protected]> wrote:
>
> On Jun 13, 2022, at 3:38 PM, Axel Rasmussen <[email protected]> wrote:
>
> > On Mon, Jun 13, 2022 at 3:29 PM Peter Xu <[email protected]> wrote:
> >> On Mon, Jun 13, 2022 at 02:55:40PM -0700, Andrew Morton wrote:
> >>> On Wed, 1 Jun 2022 14:09:47 -0700 Axel Rasmussen <[email protected]> wrote:
> >>>
> >>>> To achieve this, add a /dev/userfaultfd misc device. This device
> >>>> provides an alternative to the userfaultfd(2) syscall for the creation
> >>>> of new userfaultfds. The idea is, any userfaultfds created this way will
> >>>> be able to handle kernel faults, without the caller having any special
> >>>> capabilities. Access to this mechanism is instead restricted using e.g.
> >>>> standard filesystem permissions.
> >>>
> >>> The use of a /dev node isn't pretty. Why can't this be done by
> >>> tweaking sys_userfaultfd() or by adding a sys_userfaultfd2()?
> >
> > I think for any approach involving syscalls, we need to be able to
> > control access to who can call a syscall. Maybe there's another way
> > I'm not aware of, but I think today the only mechanism to do this is
> > capabilities. I proposed adding a CAP_USERFAULTFD for this purpose,
> > but that approach was rejected [1]. So, I'm not sure of another way
> > besides using a device node.
> >
> > One thing that could potentially make this cleaner is, as one LWN
> > commenter pointed out, we could have open() on /dev/userfaultfd just
> > return a new userfaultfd directly, instead of this multi-step process
> > of open /dev/userfaultfd, NEW ioctl, then you get a userfaultfd. When
> > I wrote this originally it wasn't clear to me how to get that to
> > happen - open() doesn't directly return the result of our custom open
> > function pointer, as far as I can tell - but it could be investigated.
>
> If this direction is pursued, I think that it would be better to set it as
> /proc/[pid]/userfaultfd, which would allow remote monitors (processes) to
> hook into userfaultfd of remote processes. I have a patch for that which
> extends userfaultfd syscall, but /proc/[pid]/userfaultfd may be cleaner.
Hmm, one thing I'm unsure about -
If a process is able to control another process' memory like this,
then this seems like exactly what CAP_SYS_PTRACE is intended to deal
with, right? So I'm not sure this case is directly related to the one
I'm trying to address.
This also seems distinct to me versus the existing way you'd do this,
which is open a userfaultfd and register a shared memory region, and
then fork(). Now you can control your child's memory with userfaultfd.
But, attaching to some other, previously-unrelated process with
/proc/[pid]/userfaultfd seems like a clear case for CAP_SYS_PTRACE.
>
On Tue, Jun 14, 2022 at 12:09 PM Peter Xu <[email protected]> wrote:
>
> On Wed, Jun 01, 2022 at 02:09:47PM -0700, Axel Rasmussen wrote:
> > Historically, it has been shown that intercepting kernel faults with
> > userfaultfd (thereby forcing the kernel to wait for an arbitrary amount
> > of time) can be exploited, or at least can make some kinds of exploits
> > easier. So, in 37cd0575b8 "userfaultfd: add UFFD_USER_MODE_ONLY" we
> > changed things so, in order for kernel faults to be handled by
> > userfaultfd, either the process needs CAP_SYS_PTRACE, or this sysctl
> > must be configured so that any unprivileged user can do it.
> >
> > In a typical implementation of a hypervisor with live migration (take
> > QEMU/KVM as one such example), we do indeed need to be able to handle
> > kernel faults. But, both options above are less than ideal:
> >
> > - Toggling the sysctl increases attack surface by allowing any
> > unprivileged user to do it.
> >
> > - Granting the live migration process CAP_SYS_PTRACE gives it this
> > ability, but *also* the ability to "observe and control the
> > execution of another process [...], and examine and change [its]
> > memory and registers" (from ptrace(2)). This isn't something we need
> > or want to be able to do, so granting this permission violates the
> > "principle of least privilege".
> >
> > This is all a long winded way to say: we want a more fine-grained way to
> > grant access to userfaultfd, without granting other additional
> > permissions at the same time.
> >
> > To achieve this, add a /dev/userfaultfd misc device. This device
> > provides an alternative to the userfaultfd(2) syscall for the creation
> > of new userfaultfds. The idea is, any userfaultfds created this way will
> > be able to handle kernel faults, without the caller having any special
> > capabilities. Access to this mechanism is instead restricted using e.g.
> > standard filesystem permissions.
> >
> > Signed-off-by: Axel Rasmussen <[email protected]>
> > ---
> > fs/userfaultfd.c | 76 ++++++++++++++++++++++++++------
> > include/uapi/linux/userfaultfd.h | 4 ++
> > 2 files changed, 66 insertions(+), 14 deletions(-)
> >
> > diff --git a/fs/userfaultfd.c b/fs/userfaultfd.c
> > index e943370107d0..8b92c1398169 100644
> > --- a/fs/userfaultfd.c
> > +++ b/fs/userfaultfd.c
> > @@ -30,6 +30,7 @@
> > #include <linux/security.h>
> > #include <linux/hugetlb.h>
> > #include <linux/swapops.h>
> > +#include <linux/miscdevice.h>
> >
> > int sysctl_unprivileged_userfaultfd __read_mostly;
> >
> > @@ -413,13 +414,8 @@ vm_fault_t handle_userfault(struct vm_fault *vmf, unsigned long reason)
> >
> > if (ctx->features & UFFD_FEATURE_SIGBUS)
> > goto out;
> > - if ((vmf->flags & FAULT_FLAG_USER) == 0 &&
> > - ctx->flags & UFFD_USER_MODE_ONLY) {
> > - printk_once(KERN_WARNING "uffd: Set unprivileged_userfaultfd "
> > - "sysctl knob to 1 if kernel faults must be handled "
> > - "without obtaining CAP_SYS_PTRACE capability\n");
> > + if (!(vmf->flags & FAULT_FLAG_USER) && (ctx->flags & UFFD_USER_MODE_ONLY))
> > goto out;
> > - }
> >
> > /*
> > * If it's already released don't get it. This avoids to loop
> > @@ -2052,19 +2048,33 @@ static void init_once_userfaultfd_ctx(void *mem)
> > seqcount_spinlock_init(&ctx->refile_seq, &ctx->fault_pending_wqh.lock);
> > }
> >
> > -SYSCALL_DEFINE1(userfaultfd, int, flags)
> > +static inline bool userfaultfd_allowed(bool is_syscall, int flags)
> > +{
> > + bool kernel_faults = !(flags & UFFD_USER_MODE_ONLY);
> > + bool allow_unprivileged = sysctl_unprivileged_userfaultfd;
> > +
> > + /* userfaultfd(2) access is controlled by sysctl + capability. */
> > + if (is_syscall && kernel_faults) {
> > + if (!allow_unprivileged && !capable(CAP_SYS_PTRACE))
> > + return false;
> > + }
> > +
> > + /*
> > + * For /dev/userfaultfd, access is to be controlled using e.g.
> > + * permissions on the device node. We assume this is correctly
> > + * configured by userspace, so we simply allow access here.
> > + */
> > +
> > + return true;
> > +}
>
> This helper reads a bit weird because potentially it constantly returns
> "true" for !syscall use case but it's very not obvious..
>
> Would it be cleaner to not pass in the bool at all? Something like (I also
> un-nested some of the condition checks, hopefully it'll be easier to read):
>
> bool userfaultfd_syscall_allowed(int flags)
> {
> /* Userspace-only page faults are always allowed */
> if (flags & UFFD_USER_MODE_ONLY)
> return true;
>
> /*
> * The user is requesting kernel fault capabilities. Privileged
> * users are always allowed even for kernel fault traps.
> */
> if (capable(CAP_SYS_PTRACE))
> return true;
>
> /* Whether we allow unprivileged users for kernel faults? */
> return sysctl_unprivileged_userfaultfd;
> }
>
> Then below...
>
> > +
> > +static int new_userfaultfd(bool is_syscall, int flags)
> > {
> > struct userfaultfd_ctx *ctx;
> > int fd;
> >
> > - if (!sysctl_unprivileged_userfaultfd &&
> > - (flags & UFFD_USER_MODE_ONLY) == 0 &&
> > - !capable(CAP_SYS_PTRACE)) {
> > - printk_once(KERN_WARNING "uffd: Set unprivileged_userfaultfd "
> > - "sysctl knob to 1 if kernel faults must be handled "
> > - "without obtaining CAP_SYS_PTRACE capability\n");
> > + if (!userfaultfd_allowed(is_syscall, flags))
> > return -EPERM;
> > - }
>
> .. we could write it as:
>
> if (is_syscall && !userfaultfd_syscall_allowed(flags))
> return -EPERM;
>
> What do you think?
>
> >
> > BUG_ON(!current->mm);
> >
> > @@ -2083,6 +2093,10 @@ SYSCALL_DEFINE1(userfaultfd, int, flags)
> > refcount_set(&ctx->refcount, 1);
> > ctx->flags = flags;
> > ctx->features = 0;
> > + /*
> > + * If UFFD_USER_MODE_ONLY is not set, then userfaultfd_allowed() above
> > + * decided that kernel faults were allowed and should be handled.
> > + */
>
> Hmm.. why this needs to be added above "released=false"? Did you want to
> add this (perhaps) above "flags" instead?
>
> IMHO when people reading the flags it'll be clear already on how it was
> handled, the thing is the comment probably hide deep anyway so I'd consider
> omitting it.
>
> The rest looks good to me, thanks.
Thanks for reviewing, Peter! Most of these comments look good to me,
I'll include them in a v4 after I get back to the office in about a
week.
>
> > ctx->released = false;
> > atomic_set(&ctx->mmap_changing, 0);
> > ctx->mm = current->mm;
> > @@ -2098,8 +2112,42 @@ SYSCALL_DEFINE1(userfaultfd, int, flags)
> > return fd;
> > }
> >
> > +SYSCALL_DEFINE1(userfaultfd, int, flags)
> > +{
> > + return new_userfaultfd(true, flags);
> > +}
> > +
> > +static int userfaultfd_dev_open(struct inode *inode, struct file *file)
> > +{
> > + return 0;
> > +}
> > +
> > +static long userfaultfd_dev_ioctl(struct file *file, unsigned int cmd, unsigned long flags)
> > +{
> > + if (cmd != USERFAULTFD_IOC_NEW)
> > + return -EINVAL;
> > +
> > + return new_userfaultfd(false, flags);
> > +}
> > +
> > +static const struct file_operations userfaultfd_dev_fops = {
> > + .open = userfaultfd_dev_open,
> > + .unlocked_ioctl = userfaultfd_dev_ioctl,
> > + .compat_ioctl = userfaultfd_dev_ioctl,
> > + .owner = THIS_MODULE,
> > + .llseek = noop_llseek,
> > +};
> > +
> > +static struct miscdevice userfaultfd_misc = {
> > + .minor = MISC_DYNAMIC_MINOR,
> > + .name = "userfaultfd",
> > + .fops = &userfaultfd_dev_fops
> > +};
> > +
> > static int __init userfaultfd_init(void)
> > {
> > + WARN_ON(misc_register(&userfaultfd_misc));
> > +
> > userfaultfd_ctx_cachep = kmem_cache_create("userfaultfd_ctx_cache",
> > sizeof(struct userfaultfd_ctx),
> > 0,
> > diff --git a/include/uapi/linux/userfaultfd.h b/include/uapi/linux/userfaultfd.h
> > index 7d32b1e797fb..005e5e306266 100644
> > --- a/include/uapi/linux/userfaultfd.h
> > +++ b/include/uapi/linux/userfaultfd.h
> > @@ -12,6 +12,10 @@
> >
> > #include <linux/types.h>
> >
> > +/* ioctls for /dev/userfaultfd */
> > +#define USERFAULTFD_IOC 0xAA
> > +#define USERFAULTFD_IOC_NEW _IO(USERFAULTFD_IOC, 0x00)
> > +
> > /*
> > * If the UFFDIO_API is upgraded someday, the UFFDIO_UNREGISTER and
> > * UFFDIO_WAKE ioctls should be defined as _IOW and not as _IOR. In
> > --
> > 2.36.1.255.ge46751e96f-goog
> >
>
> --
> Peter Xu
>
On Jun 14, 2022, at 5:55 PM, Axel Rasmussen <[email protected]> wrote:
> ⚠ External Email
>
> On Mon, Jun 13, 2022 at 5:10 PM Nadav Amit <[email protected]> wrote:
>> On Jun 13, 2022, at 3:38 PM, Axel Rasmussen <[email protected]> wrote:
>>
>>> On Mon, Jun 13, 2022 at 3:29 PM Peter Xu <[email protected]> wrote:
>>>> On Mon, Jun 13, 2022 at 02:55:40PM -0700, Andrew Morton wrote:
>>>>> On Wed, 1 Jun 2022 14:09:47 -0700 Axel Rasmussen <[email protected]> wrote:
>>>>>
>>>>>> To achieve this, add a /dev/userfaultfd misc device. This device
>>>>>> provides an alternative to the userfaultfd(2) syscall for the creation
>>>>>> of new userfaultfds. The idea is, any userfaultfds created this way will
>>>>>> be able to handle kernel faults, without the caller having any special
>>>>>> capabilities. Access to this mechanism is instead restricted using e.g.
>>>>>> standard filesystem permissions.
>>>>>
>>>>> The use of a /dev node isn't pretty. Why can't this be done by
>>>>> tweaking sys_userfaultfd() or by adding a sys_userfaultfd2()?
>>>
>>> I think for any approach involving syscalls, we need to be able to
>>> control access to who can call a syscall. Maybe there's another way
>>> I'm not aware of, but I think today the only mechanism to do this is
>>> capabilities. I proposed adding a CAP_USERFAULTFD for this purpose,
>>> but that approach was rejected [1]. So, I'm not sure of another way
>>> besides using a device node.
>>>
>>> One thing that could potentially make this cleaner is, as one LWN
>>> commenter pointed out, we could have open() on /dev/userfaultfd just
>>> return a new userfaultfd directly, instead of this multi-step process
>>> of open /dev/userfaultfd, NEW ioctl, then you get a userfaultfd. When
>>> I wrote this originally it wasn't clear to me how to get that to
>>> happen - open() doesn't directly return the result of our custom open
>>> function pointer, as far as I can tell - but it could be investigated.
>>
>> If this direction is pursued, I think that it would be better to set it as
>> /proc/[pid]/userfaultfd, which would allow remote monitors (processes) to
>> hook into userfaultfd of remote processes. I have a patch for that which
>> extends userfaultfd syscall, but /proc/[pid]/userfaultfd may be cleaner.
>
> Hmm, one thing I'm unsure about -
>
> If a process is able to control another process' memory like this,
> then this seems like exactly what CAP_SYS_PTRACE is intended to deal
> with, right? So I'm not sure this case is directly related to the one
> I'm trying to address.
>
> This also seems distinct to me versus the existing way you'd do this,
> which is open a userfaultfd and register a shared memory region, and
> then fork(). Now you can control your child's memory with userfaultfd.
> But, attaching to some other, previously-unrelated process with
> /proc/[pid]/userfaultfd seems like a clear case for CAP_SYS_PTRACE.
I agree about CAP_SYS_PTRACE. I just know that if the /dev approach is
taken, there would be even more pushback for userfaultfd2.
Whatever.